Ontology meeting 2014-08-28
Attendees: DavidH, Harold, Heiko, Tanya, Chris, PaulT
- 1 Agenda
- 1.1 GO.curator_dbxrefs freeze
- 1.2 SO Imports
- 1.3 PR Imports
- 1.4 OBA Imports
- 1.5 Assert Report
- 1.6 Proposal: verification check to ensure certain term patterns are axiomatized
- 1.7 Improving TermGenie accessibility to templates
- 1.8 Annotation relations
- 1.9 Revisit MF refactoring?
- 1.10 Mapping extensions to LEGO
At some point in the near future we'd like to place a freeze on GO.curator_dbxrefs editing.
Mostly the ontology group that edits this: http://viewvc.geneontology.org/viewvc/GO-SVN/trunk/doc/GO.curator_dbxrefs?view=log
We will switch to using a yaml file as the source version (we will still provide a read-only version for legacy apps).
The version that is consumed by most apps will be JSON, similar to this: https://s3.amazonaws.com/go-public/metadata/users.json
The metadata will be minimal - it's deigned to be JOINed with other sources of researcher data. As an end to this we will be encouraging use 3rd party URLs as identifiers, e.g. orcid, although GOC IDs will be supported legacy.
Freeze coming soon.Edits going forward will be done on the yaml file which will likely be @ github. If possible, have each person be uniquely identified by an external resource, like ORCID. However, GOC ids (GOC:dph) will still be allowed. We'll continue to use Persona for logging into TermGenie and Noctua.
They're in the live file. Problem: currently no links between SO and ChEBI and sometimes we reference SO and sometimes we use ChEBI in the xps. This can lead to incomplete reference. At some point, we will autoreplace SO with MSO and that will have join points to ChEBI. SO and ChEBI will need to sort out who has what class.
Action item: Look ok?
- https://sourceforge.net/p/geneontology/ontology-requests/11092/ - Harold to report
PR imports added, some problems resolved macromolecular complex (GO) is SubClass of molecular entity (ChEBI), imports made more minimal, only importing PR ids and the subsets necessary for closure. Eventually will import links to the upper ontologies for better reasoning.
Found lots of overlap between ChEBI and GO, things that need resolution. Discussion begun with ChEBI. For cases where there are terms with the same strings in two different ontologies. Ideally, one ontology would yield but if not other agreements could be made.
- When do we add new complexes to the ontology. Currently there are two requests in the TG queue that differ only by subunit composition. Shouldn't these now be handled by PR?
Accept for now, cross the bridge of adding every complex with varying composition if/when we reach it.
Note in some cases we have go->oba->go axioms. oba is acting as a shortcut ontology for talking about things in the GO
GO:0044375 ! regulation of peroxisome size OBA:0000122 ! peroxisome size GO:0005777 ! peroxisome GO:0060078 ! regulation of postsynaptic membrane potential OBA:0000125 ! post-synaptic membrane potential GO:0045211 ! postsynaptic membrane GO:0051563 ! smooth endoplasmic reticulum calcium ion homeostasis OBA:0000043 ! calcium ion concentration in smooth ER GO:0005790 ! smooth endoplasmic reticulum GO:0032468 ! Golgi calcium ion homeostasis OBA:0000005 ! Golgi calcium ion concentration GO:0005794 ! Golgi apparatus
After getting these in, we will have linked out to all the basic external ontologies that we need. Chris will do a trial run in adding these in, will check for strange inferences, then will commit.
Anything to be done or wait for DOS?
DH will get rid of part of relationships that are in that report.
Proposal: verification check to ensure certain term patterns are axiomatized
Add a jenkins check to ensure that terms that follow a certain lexical pattern must have equivalence axioms. But we would need some kind of flag to subvert this as there will be exceptions
Improving TermGenie accessibility to templates
Prototype approach - link out from term detail page, from AmiGO perhaps or from term completion Documentation approach - may be more involved
File still needs to be converted into owl.
Revisit MF refactoring?
Where do we stand in that process? - Paul T. Some comments made via WebProtege, DavidOS set this up. Paul responded to David's comments. EBI folks are away right now. Wait for DOS to return for update. Prototype Noctua using the refactored ontology? Minimal steps were outlined in document. Can this be done? No time to revisit yet. Next step: Noctua meeting - next ontology slot?
Mapping extensions to LEGO
For consistency and translation, progress? For each extension + annotation relation, need a NOCTUA diagram and provide a link in the documentation. Simulating what would be done as annotators. http://wiki.geneontology.org/index.php/Category:Annotation_extension Make an NOCTUA example for each, some are trivial (occurs_in), some are not, examples: dependent_on, causally_upstream_of, any where the filler for the relation is a protein or a complex. Chris will tag the ones that need NOCTUA diagrams. PaulT will contribute too! Goal is to be done by Barcelona. GO eds to discuss at next mtg.